Summary

  • Tomohiro Fujisaki and Ruri Hiromi's first APNIC prop-115 draft gave two homes the same IPv4 address and different port ranges; it also sought more specific IPv6 assignment information for filtering decisions.
  • Drafts two and three put port ranges and IPv6 prefix sizes into the Whois proposal. Draft four dropped the port-range idea, kept the IPv6 problem and listed five possible places to publish the answer. The proposal was withdrawn before APNIC 41.

An IP address was once a decent clue to where traffic came from. Put two customers behind the same address with carrier-grade NAT, and the clue needs a second line: a port range. The first draft of APNIC prop-115, coauthored by Tomohiro Fujisaki and Ruri Hiromi, made that problem memorable with two homes sharing one IPv4 address. Home A got one run of ports; Home B got another. The draft also asked for the size of IPv6 assignments. An operator considering a filter could then see a customer-sized boundary rather than block an ISP-sized range.

The proposed address book kept taking on jobs. In version three, the coauthors spelled out port-range information for shared IPv4 and assignment-prefix size for IPv6. The document acknowledged the price: extra registration work, possible Whois changes, operator cooperation and the question of who should see fine-grained records. A field that helps someone avoid a blunt filter still needs a maintainer whenever the underlying assignment changes. A stale port map could make the new precision theatrical.

The fourth draft tells a different story. The IPv4 port-range request is gone. IPv6 assignment size remains, but Whois is now one of five possible carriers alongside DNS, the Internet Routing Registry, another database and even the routing system. The solution section opens with an admission in plain draft language: the solution still needed specifying. The registry directory had gone from proposed answer to one candidate for the job.

The APNIC proposal history records two failed consensus attempts, at APNIC 39 and 40. The coauthors submitted version four to the chairs in February 2016, then withdrew it before it reached the mailing list. At APNIC 41, Hiromi discussed the issue in an informational presentation; no policy proposal went to that Policy SIG. Fujisaki's coauthored experiment left a useful paper trail: first the port number, then the cost of deciding where that number belongs, then a narrower question still searching for a home.